Logics of Critical Explanation in Social and Political Theory is an exceptionally valuable
contribution to political and social theory from a poststructuralist perspective. Erudite,
balanced, and supported by revealing case-studies, the book explores the inner logic of a critical
approach and places it convincingly amidst a rich tapestry of methods and strategies. Political
scientists, political theorists, and philosophers of the social sciences will find that it enhances
their own understanding of the discipline; students and general readers will be indebted to it
for a highly accessible and lucid analysis. If anyone doubted the coming of age of poststructural
political theory, here is the proof of its maturity.
Michael Freeden,
Professor of Politics,
Oxford University
This is a remarkable book, not only due to its rigour and scope, but also because of its sustained
effort to engage with the most important contemporary currents in social explanation and
social theory. As for the highly original theoretical approach that the authors present, I can
only say that it is the most significant attempt so far at elaborating a general framework for
social research from a poststructuralist perspective. The notion of logics, as defended in this text,
is bound to have a lasting influence on the work of social and political scientists.
Ernesto Laclau
Professor of Humanities
and Rhetorical Studies,
Northwestern University, USA
and Professor of Politics,
Department of Government,
University of Essex, UK
Logics of Critical Explanation in Social and Political Theory examines positivist, hermeneutic,
rational choice and other methods, drawing selective sustenance from several to build a theory of
critical explanation reducible to none. The result is a unique book in the philosophy of the social
sciences, one that appreciates an element of chanciness in the world as it retains the ambition to
explain. An indispensable book for those who are dissatisfied with the options available today.
William E. Connolly
Professor of Political Science,
Johns Hopkins University, USA
An ambitious and inventive intervention in debates on the methodology of the social sciences.
Drawing on the resources of contemporary philosophy of science as well as hermeneutic and
poststructuralist philosophy, Glynos and Howarth offer a provocative, but well-grounded,
argument for a focus on ‘logics’ as basic to the tasks of both explanantion and criticism. Their
argument is insightful, well illustrated, and should be widely debated.
David Owen
Professor of Social and Political Philosophy,
University of Southampton, UK
Glynos-FM.qxd 13/8/07 6:13 PM Page i

In this path-breaking and commanding text, Glynos and Howarth call for and set out a
distinctive approach to a reconstruction of social science. Established and emergent traditions
of enquiry – naturalism and hermeneutics – are carefully reviewed and critiqued as a vehicle
for elaborating the first systematic and sustained application of poststructuralist insights to a
critical analysis of the nature of explanation and method. This is an important, challenging
book for researchers across the social sciences, including the fields of management and business.
It also has wider relevance for anyone who uses social scientific data, or is curious about how a
‘science’ of the social might be possible.
Hugh Willmott
Professor in Organizational Studies,
University of Cardiff, UK
This edifying book is a tremendous accomplishment, offering a fresh look at both epistemology
(what counts as knowledge) and ontology (our underlying presuppositions about the way the
world goes round). But more than that, this book raises poststructuralist discourse theory to a
new level of intellectual prestige. The raw material for a coherent poststructuralist discourse
theory has been available for some time. And now, thanks to David Howarth and Jason Glynos,
the wide ranging poststructuralist literature has been deployed to redescribe social science
research as problematization and articulation – a context-sensitive research protocol that
challenges positivism’s universalizing causal laws, which have inappropriately colonized the
social sciences. This is careful yet imaginative scholarship, worthy of widespread attention
across the social sciences.
Hugh T. Miller
Professor of Public Administration
Florida Atlantic University, USA
This is one of those exceedingly rare books which is both philosophically astute as well as of
considerable practical use. It breaks new, creative ground concerning age-old issues in political
theory and the philosophy of the social sciences. At the same time, it offers to the political
analyst valuable heuristic strategies for engaging in critical empirical inquiry.
Henk Wagenaar
Professor of Public Policy
University of Leiden, Amsterdam
Beginning with a much-needed critique of scientism and the current state of affairs in the social
and political sciences, Glynos and Howarth develop an original, sophisticated and rigorous
account of discourse theory. Using the guiding idea of ‘logics of critical explanation’, Glynos
and Howarth propose nothing less than a new ontology for social and political theory. This is
excellent work that should be debated and discussed in the years to come.
Simon Critchley
Professor of Philosophy
New School for Social Research, USA
Glynos-FM.qxd 13/8/07 6:13 PM Page ii

The Social Science Wars have precipitated a renewed interest in the character,
purpose and methods of social science. Positivists and naturalists are criticized
by interpretivists and critical theorists, while quantitative researchers are
challenged by those who favour qualitative and ethnographic techniques.
In turn, mainstream social scientists have responded with a vigorous defence
and restatement of their commitments.
Logics of Critical Explanation in Social and Political Theory proposes a novel
approach to practising social and political analysis based on the role of logics.
The authors articulate a distinctive perspective on social science explanation
that avoids the problems of scientism and subjectivism by steering a careful
course between law-like explanations and thick descriptions. Drawing upon
hermeneutics, poststructuralism, psychoanalysis, and post-analytical
philosophy, this new approach offers a particular set of logics – social,
political, and fantasmatic – with which to construct critical explanations
of practices and regimes.
While the first part of the book critically engages with law-like, interpretivist
and causal approaches to critical explanation, the second part elaborates an
alternative grammar of concepts informed by an ontological stance rooted in
poststructuralist theory. In developing this approach, a number of empirical
cases are included to illustrate its basic concepts and logics, ranging from the
apartheid regime in South Africa to recent changes in higher education.
The book will be a valuable tool for scholars and researchers in a variety
of related fields of study in the social sciences, especially the disciplines of
political science and political theory, international relations, social theory,
cultural studies, anthropology and philosophy.
Jason Glynos is a Lecturer in Political Theory in the Department of
Government at the University of Essex, UK. He is also Director of the Masters
Programme in Ideology and Discourse Analysis at the University of Essex.
David Howarth is a Senior Lecturer in Political Theory in the Department
of Government at the University of Essex, UK. He is also Co-Director of the
Centre for Theoretical Studies and Director of the Masters Programme in
Political Theory at the University of Essex.
Logics of Critical Explanation in
Social and Political Theory
Glynos-FM.qxd 13/8/07 6:13 PM Page iii

Routledge Innovations in Political Theory
1 A Radical Green Political
Theory
Alan Carter
2 Rational Woman
A feminist critique of dualism
Raia Prokhovnik
3 Rethinking State Theory
Mark J. Smith
4 Gramsci and Contemporary
Politics
Beyond pessimism of
the intellect
Anne Showstack Sassoon
5 Post-Ecologist Politics
Social theory and
the abdication of
the ecologist paradigm
Ingolfur Blühdorn
6 Ecological Relations
Susan Board
7 The Political Theory of
Global Citizenship
April Carter
8 Democracy and National
Pluralism
Edited by Ferran Requejo
9 Civil Society and Democratic
Theory
Alternative voices
Gideon Baker
10 Ethics and Politics in
Contemporary Theory
Between critical theory and
post-marxism
Mark Devenney
11 Citizenship and Identity
Towards a new republic
John Schwarzmantel
12 Multiculturalism, Identity
and Rights
Edited by Bruce Haddock and
Peter Sutch
13 Political Theory of
Global Justice
A cosmopolitan case for
the World State
Luis Cabrera
14 Democracy, Nationalism and
Multiculturalism
Edited by Ramón Maiz and
Ferrán Requejo
15 Political Reconciliation
Andrew Schaap
Glynos-FM.qxd 13/8/07 6:13 PM Page iv

16 National Cultural
Autonomy and
Its Contemporary Critics
Edited by Ephraim Nimni
17 Power and Politics in
Poststructuralist Thought
New theories of the political
Saul Newman
18 Capabilities Equality
Basic issues and problems
Edited by Alexander Kaufman
19 Morality and Nationalism
Catherine Frost
20 Principles and Political
Order
The challenge of diversity
Edited by Bruce Haddock,
Peri Roberts and Peter Sutch
21 European Integration and the
Nationalities Question
Edited by John McGarry and
Michael Keating
22 Deliberation, Social Choice
and Absolutist Democracy
David van Mill
23 Sexual Justice/Cultural
Justice
Critical perspectives in political
theory and practice
Edited by Barbara Arneil,
Monique Deveaux, Rita Dhamoon
and Avigail Eisenberg
24 The International Political
Thought of Carl Schmitt
Terror, liberal war and
the crisis of global order
Edited by Louiza Odysseos and
Fabio Petito
25 In Defense of Human Rights
A non-religious grounding in a
pluralistic world
Ari Kohen
26 Logics of Critical Explanation
in Social and Political Theory
Jason Glynos and David Howarth
Glynos-FM.qxd 13/8/07 6:13 PM Page v

Glynos-FM.qxd 13/8/07 6:13 PM Page vi

Logics of Critical Explanation
in Social and Political Theory
Jason Glynos and David Howarth
Glynos-FM.qxd 13/8/07 6:13 PM Page vii

First published 2007
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
Simultaneously published in the USA and Canada
by Routledge
270 Madison Ave, New York, NY 10016
Routledge is an imprint of the Taylor & Francis Group,
an informa business
© 2007 Jason Glynos and David Howarth
All rights reserved. No part of this book may be reprinted or
reproduced or utilised in any form or by any electronic,
mechanical, or other means, now known or hereafter
invented, including photocopying and recording, or in any
information storage or retrieval system, without permission in
writing from the publishers.
British Library Cataloguing in Publication Data
A catalogue record for this book is available
from the British Library
Library of Congress Cataloging in Publication Data
A catalog record for this book has been requested
ISBN10: 0–415–40428–2 (hbk)
ISBN10: 0–203–93475–X (ebk)
ISBN13: 978–0–415–40428–0 (hbk)
ISBN13: 978–0–203–93475–3 (ebk)
Glynos-FM.qxd 13/8/07 6:13 PM Page viii
This edition published in the Taylor & Francis e-Library, 2007.
“To purchase your own copy of this or any of Taylor & Francis or Routledge’s
collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.”
ISBN 0-203-93475-X Master e-book ISBN

Contents
Acknowledgements xi
Introduction 1
1 Retroduction 18
2 Contextualized self-interpretations 49
3 Causal mechanisms 83
4 Ontology 103
5 Logics 133
6 Articulation 165
Conclusion 209
Notes 216
Bibliography 237
Index 253
Glynos-FM.qxd 13/8/07 6:13 PM Page ix

Glynos-FM.qxd 13/8/07 6:13 PM Page x

As Michel Foucault often noted, writing a book is best practised as an
intellectual adventure, whose final destination and ultimate outcome can
never be guaranteed. It involves intellectual collaboration, a fair bit of financial
and emotional support, not to mention the good graces of Fortuna. In all these
respects, this book is no exception.
First, and foremost, it is the result of an intense intellectual collaboration
between the two authors, an intensity that was noted with some curiosity, and
perhaps not a little amusement, by administrators and staff in the Literature
and Film Studies Department at the University of Essex, whose reading room
we often colonized when trying to render our often inchoate thoughts a little
more coherent. The arguments that have emerged from this collaboration
reflect our ongoing engagement with different theoretical traditions, as well
as different approaches and methods in social science and political analysis,
which we have been teaching, researching, and discussing since our appointment
to the Department of Government several years ago.
In particular, our ideas have been powerfully shaped by our respective
encounters in the Ideology and Discourse Analysis (IDA) Programme in
the Department of Government, also known as the Essex School in Discourse
Theory. The IDA programme was established by Ernesto Laclau shortly
after the publication of Hegemony and Socialist Strategy in 1985 (Laclau and
Mouffe, 1985). Ernesto has been a constant source of inspiration and support
for the ideas and arguments developed in this book. In equal fashion, the
ideas put forward here also bear the strong imprint of other core members of
this research programme, especially Aletta Norval and Yannis Stavrakakis.
In this context, we should also mention Lasse Thomassen and Jacob Torfing,
from whom we received much encouragement for our project, and who
also read several drafts of individual chapters and even versions of entire
manuscripts.
Many of those in the programme with whom we have discussed the issues
explored in this book, and from whom we have profited, are now academics
teaching and researching in universities and institutes across the globe. We
have benefited from their input partly as a result of the recently established
Ideology and Discourse Analysis World Network (IDA World), whose aim is
Acknowledgements
Glynos-FM.qxd 13/8/07 6:13 PM Page xi

to facilitate the exchange of ideas on the topic of IDA through the internet,
as well as through conferences and annual workshops. In this context, we
have benefited from the feedback of Sebastian Barros, Mercedes Barros,
Mark Devenney, Torben Dyrberg, Alex Groppo, Alan Hansen, Juan-Pablo
Lichtmeyer, Oliver Marchart, Emilia Palonen, Carlos Pessoa, Anna Schrober,
Rei Shigeno, and Rosa Nidia Buenfil.
Earlier drafts also enjoyed the discerning gaze of members of the political
theory group at Essex over the course of the last few years. These include
Sheldon Leader, Albert Weale, and Richard Bellamy. Others here at Essex and
far beyond have also offered us helpful feedback in the context of numerous
conferences, workshops, and summer schools around the world, including
Steven Griggs, Sean Nixon, Mike Roper, Eva Sorenson, Tamara Metze,
Margo van den Brink, Peter Kitchenman, and Todd Bridgman. Their
comments have been most useful in helping us to express our arguments in
a clearer fashion.
Three institutions in particular have provided the main arenas for many of
the collaborations and encounters mentioned thus far: the Doctoral Research
Programme in IDA, the Centre for Theoretical Studies, and the Department
of Government, all at the University of Essex. The flourishing Doctoral
Research Programme has provided an outstanding context within which to
discuss theoretical, methodological, and empirical research. Our book has
drawn inspiration from, and has been motivated in part by, our individual
and collective interactions with our Masters and PhD students over the last
five years or so, too numerous to name in person, but no less important for that.
But our book was also discussed in detail over many PhD seminars during the
course of the last academic year, each chapter benefiting considerably from
this public exposure and perceptive critique. These encounters embodied for
us genuine collective research at its best, helping us to shape our thoughts
into a more robust final product. Our special thanks in this regard go to those
who attended this seminar series, comprising both IDA PhD students and
visiting PhD students: Pete Bloom, Ryan Brading, Carl Cederstrom,
Sam Dallyn, Erdem Damar, Peter Edwards, Jonathan Dean, Laura Glanc,
Steve Gormley, Jenny Gunnarsson-Payne, Cengiz Gunes, Sarah Hartley,
Christos Iliadis, Leonidas Karakatsanis, Stefan Milizer, David Payne,
Isis Sanchez, and Wei Yan Chang.
In our experience, the University of Essex has always been at the centre of
passionate intellectual engagement and discussion, and the Centre for
Theoretical Studies has added its not inconsiderable weight and resource to
this flourishing intellectual environment. Not only has it served to connect
different sorts of critical theorizing within the university, but it has also
enabled us to witness and participate in some of the leading intellectual
debates and conversations in the contemporary human and social sciences,
many of which have shaped the writing of this book. In this regard, Slavoj
i˚ek’s long association with the Centre and the IDA programme has been a
source of much intellectual nourishment. Of equal import, from our point of
xii Acknowledgements
Glynos-FM.qxd 13/8/07 6:13 PM Page xii

view, have been the debates between poststructuralists and post-Marxists on
the one hand, and critical realists on the other. In addition, as graduate stu-
dents and lecturers, we have been fortunate to attend lectures and mini-
courses by the likes of Bill Connolly, Jacques Derrida, Hubert Dreyfus,
Stephen Mulhall, James Tully, Rudi Visker, and Hugh Willmott, and traces
of their work will no doubt be discerned in our book.
The Department of Government at the University of Essex continues to
provide a demanding and rigorous institutional context in which to converse
and debate politics and political analysis. In addition to those we have already
mentioned, we are grateful to David Sanders and Carole Parmenter for their
supportive stewardship of the department. We are also grateful for the constant
encouragement and friendly criticism of our colleagues John Bartle and
Joe Foweraker, as well as Todd Landman, who engaged us in debates about
method from a very early stage. Finally, the Essex Discourse Theory Summer
School, which is hosted by the Department of Government, has provided us
with the opportunity to develop some of the ideas presented in this book. In
particular, we would like to thank Elinor Scarborough and Eric Tanenbaum
for encouraging this approach to social and political analysis in the Summer
School as a whole, as well as the many students who participated in these
courses and seminars for their interesting questions and views.
Lastly, but certainly not least, we would like to thank Jackie Pells for her
efficient and cheerful administrative support in helping us to complete this
manuscript. In this regard, too, we would like to thank Heidi Bagtazo,
Harriet Brinton, and Amelia McLaurin at Routledge for their patience,
expertise, and goodwill in overseeing the publication of this book. Thanks
finally to the anonymous reviewers of our original manuscript.
It is clear that our book is deeply indebted to the many people who have
inspired and critically engaged with us, as well as to the University of Essex,
which has provided much more than one would expect from a campus
university in the northeast Essex countryside. But, of course, the real debt is to
our families, who have supported us through thick and thin. We thus dedicate
this book to our families: Anita, Michael, Byron, Jes, Jordan, Carter, Moe, Ivan,
Eric, and Lorraine; and Aletta and James.
Colchester, 9 March 2007
Acknowledgements xiii
Glynos-FM.qxd 13/8/07 6:13 PM Page xiii

Glynos-FM.qxd 13/8/07 6:13 PM Page xiv

In the final analysis we are not confronted with exclusive choices: either
empirical theory or interpretative theory or critical theory. Rather, there is an
internal dialectic in the restructuring of social and political theory. When we
work through any of these moments, we discover how the others are
implicated. An adequate social and political theory must be empirical,
interpretative, and critical.
(Bernstein 1976: 235)
More than thirty years has now passed since Richard Bernstein wrote these
concluding words to The Restructuring of Social and Political Theory. Having
criticized the then – and still – dominant view that social and political theorists
ought to distinguish and then choose between explanation, interpretation
and critique, he optimistically advocates their dialectical integration for any
theoretical approach worth its salt. In the meantime, the contours of our
intellectual landscapes in the social sciences, not to mention the political and
cultural landscapes, have changed considerably. The fashionable interest in
Marxism, linguistic philosophy, existentialism, and the Frankfurt school of
critical theory, has given way to discussions about postmodernism, critical
realism, interpretivism, poststructuralism, deconstruction, post-colonialism,
and so forth.
But one thing has remained constant during this period, and that is the
extraordinary resilience of positivism in social and political studies, whose
protagonists desire a fully-fledged scientific study of politics and society.1
Indeed, though one may point to important challenges facing this onward
march of ‘positive theory’, its overall trajectory and momentum appear diffi-
cult to check. For instance, while the emergence of the ‘perestroika movement’
in political science,2
coupled with a renewed interest in various interpretive
approaches, are welcome interventions in the field of political analysis, it is
important to ensure that they represent more than just a momentary backlash
that would only confirm the marginalization of critical and interpretive
approaches.
Introduction
Glynos-Intro.qxd 10/8/07 12:47 PM Page 1

2 Introduction
An underlying assumption of this book is that unless this unilateral
forward march is checked, Bernstein’s hope for an integrated social and political
theory, which does not have to choose between explanation, interpretation
and critique, will remain an empty dream. For this reason, we seek to reactivate
a crisis in the social sciences. Here, we allude to Edmund Husserl’s diagnosis
of ‘a crisis of the European sciences’ in the 1930s, in which the founder of
transcendental phenomenology felt his entire intellectual project threatened
by a growing philosophical naiveté that was crystallized either in an unreflective
objectivism and narrow-minded positivism, or a relapse into forms of
irrationalism and subjectivism (see Husserl 1965: 179–85).
Talk of a crisis in the social sciences, especially in the fields of social and
political analysis, is perhaps an exaggeration, or even a cliché. Nevertheless,
the demands for greater methodological pluralism emanating from certain
quarters of the American Political Science Association, renewed debates
about the character and purpose of social and political analysis in Britain and
the rest of Europe, the growth of new or revamped approaches to social
science research (e.g. feminism, discourse analysis, psychoanalysis, critical
realism, hermeneutics, critical theory, deconstruction, and so forth), together
with criticisms about the inadequacies of the social sciences in elucidating
and explaining, let alone predicting, important changes in our rapidly
changing world, attest to a growing unease about the methods, purposes and
ideals of social science.3
Scientism and beyond
One way to conceptualize this emergent disquiet is to highlight the increas-
ing scientism of the dominant approaches and methods of social and political
theory. To use Habermas’s prescient formulation, this disquiet is focused on
the widespread ‘conviction that we can no longer understand science as one
form of knowledge, but rather must identify knowledge with science’
(Habermas 1978: 4). For us, the chief problem with this underlying disposition
in social and political analysis is the pre-dominance of an elusive and
unattainable ideal: a science of politics and society – at least one modelled on
a particular conception of natural science. This ideal, which stretches back at
least to Hobbes, has not only proved difficult to realize in the form of laws
and reliable empirical generalizations (whether causal or correlational), but it
has skewed the overall purposes of the social sciences, separating positive
science from questions of critique and evaluation.
The dream of those who want a science of politics in the modern period
comprises, among other things, the discovery of a set of laws or robust
empirical generalizations that approximate those found in the natural sciences,
which would allow political scientists, as well as policymakers, administrators
and practitioners, to explain and predict relevant political events and practices.
Michel’s ‘iron law of oligarchy’ and ‘Duverger’s law’ lent early hope to the
belief that the behaviour of political parties, voters, groups and policy-makers,
Glynos-Intro.qxd 10/8/07 12:47 PM Page 2

not to mention the effects of electoral systems and other political institutions
on political activity, could all be drawn under the sway of a purely positive
theory. Over the course of the twentieth century this dream has had many
incarnations, beginning with the development of political science in the
United States at the start of the century, followed by the behaviouralist and
post-behaviouralist revolutions of the post-war period, and the growth of
formal approaches, principally in the guise of rational choice and game-
theoretical accounts of political science.4
Each, however, has been predicated
on laying the solid foundation of a positive science and thus ending
philosophical debate about the contested status of social and political science.
What is problematic with this vision is not just an empirical deficiency
that centres on the paucity of actual laws or weaknesses in predictive capabili-
ties, but also the way in which this vision is underscored by an ontological
deficiency that raises profound doubts about the very desirability of positive
theory. To develop this claim, we focus initially on the problem of prediction
and its constitutive connection to the law-like model of social science expla-
nation. Of course, the notion of prediction has acquired numerous connota-
tions. Often it is linked to the task of forecasting events and processes, whether
they are elections or economic cycles. At other times, it is equated with the
making of ‘informed conjectures’ about the possible trends or outcomes of
social and political processes. In order to clarify the way we propose to use the
concept of prediction in our book, we can appeal to Derek Sayer’s helpful
distinction between prognosis and prediction. While prognosis refers to ‘the
likely course of future events which, although well-grounded in our analysis
of the conditions and mechanisms underpinning present phenomena, cannot
be generated out of this analysis by simple deduction’, prediction is understood
as ‘a deduction of what will necessarily follow if (1) certain laws, L1...n,
themselves deducible from the theory, T, obtain, and (2) requisite antecedent
conditions, C1...n, are satisfied’ (Sayer 1983: 139). In our book we focus
mainly on the way prediction in this latter sense is deployed in relation to
processes of theory-formation and theory-testing, and thus in relation to the
explanatory capacity of scientific theories.
It is therefore clear that a key motivating factor of our book concerns the
spectre of scientism, especially its current positivist incarnation, which is evident
in the paradigmatic status accorded to causal laws. But naturally our general
dissatisfaction with the causal law paradigm also leads us to consider the
alternatives. In fact, two leading responses to the inadequacies of the law-like
model are considered: those that stress the role of contextualized self-interpretations
and those that emphasize the role of causal mechanisms. With the term ‘con-
textualized self-interpretation’ we gather together a set of hermeneutical theo-
ries which explicitly and directly question the compatibility of the natural and
social sciences, and which offer an alternative ontological framework in which
to embed their accounts of the social and political worlds. In this approach,
thick descriptions of individual and collective meanings, beliefs, and traditions
are opposed to the search for law-like explanations of social phenomena.
Introduction 3
Glynos-Intro.qxd 10/8/07 12:47 PM Page 3

But the critiques of the causal law paradigm are not confined to
hermeneutical positions. On the contrary, those who favour the use of causal
mechanisms to explain social phenomena are also sceptical of any attempt to
make causal laws and prediction constitutive of social science explanation.5
Thus, in Elster’s words, the use of mechanisms can sometimes help us to
‘explain without being able to predict, and sometimes predict without being
able to explain’. But while many scientific theories enable us to do both, this is
the ‘exception rather than the rule’ in the social sciences (Elster 1989: 8). This
view has been more strongly expressed by critical realists like Roy Bhaskar, who
feel that ‘the appraisal and development of theories in the social sciences cannot
be predictive and so must be exclusively explanatory’ (Bhaskar 1998: 21).
These two sorts of critique of the causal law paradigm are helpful in exposing
difficulties in the dominant paradigm, but they are not themselves without
difficulty. Advocates of causal mechanisms err on the side of abstracting
mechanisms from the historical contexts in which they function, thus reifying
them in a way that constrains their contingency and militates against their full
contextualization. In this approach, the universalism of the causal law ideal still
exerts a powerful attraction. Proponents of contextualized self-interpretations,
by contrast, run the risk of simply extolling the virtues of historical context and
concrete particularity, thus precluding the development of critical explanations
that can somehow transcend the particularity of a given situation, both in terms
of accounting for practices and in providing an immanent critical vantage-point
for their evaluation and political engagement.
In sum, the dilemmas that arise from our initial problematization of the
dominant approaches to social and political analysis can be expressed in
the form of two questions. Can we develop an approach that respects the self-
interpretations of social actors, while not reducing explanations to their
subjective viewpoints alone? Is it possible to have a type of explanation that
admits of a certain generality, provides the space for critique, and yet respects
the specificity of the case under investigation? In responding to these perennial
questions, which we do in the affirmative, it is necessary to develop an
approach to explanation that satisfies certain valid insights of hermeneutics
and naturalism, but which does not retreat back to a purely subsumptive or
ideographic account of social and political phenomena. For this purpose,
we develop the notion of logics, counter-posing them not simply to causal
laws but also to causal mechanisms and contextualized self-interpretations.
In so doing, we draw upon the poststructuralist tradition of thought in a way
that addresses the diremptions brought about by the positivist hegemony in
contemporary social science. Even so, it is worth noting that some of the
problems we seek to address in this book reside closer to home: that is to say,
in the very tradition from which we speak.
Two dilemmas of poststructuralist discourse theory
Over the last twenty years or so, proponents of poststructuralism have made
important advances in developing an alternative ontological standpoint with
4 Introduction
Glynos-Intro.qxd 10/8/07 12:47 PM Page 4

which to critically explain a range of phenomena. More specifically, in the
field of social and political analysis, they have begun to furnish the theoretical
means to conceptualize the character and transformation of social structures,
and to clarify the relationship between social structures, political agency and
power. They have also endeavoured to provide the conceptual resources with
which to explore the political constitution and dissolution of social identities,
while striving to account for the dynamics of human subjectivity. And finally
they have sought to move us beyond the simple critique and deconstruction
of texts, practices and institutions to offer an alternative conception of ethics
for the critical evaluation of political and moral norms and structures.
Discourse theorists working within the poststructuralist tradition of
thought – and this book is firmly situated within this subset of poststructuralist
theory – focus their attention on the reproduction and transformation of
hegemonic orders and practices.6
They develop the theoretical means to account
for the ways in which subjects are gripped by certain ideologies or discourses
(even if the latter are not necessarily in their interests, or indeed consistent
with their beliefs), while also seeking to account for the different ways in
which dominant orders are contested by counter-hegemonic or other resistance
projects, where the latter involve the construction of new identities. Of
central importance in this regard is an insistence on ‘the primacy of politics’
to explain and critically engage with a range of social phenomena. In short,
discourse theorists have developed – and are continuing to develop and
refine – the conceptual grammars with which to account for the way certain
political projects or social practices remain or become hegemonic.7
In more concrete terms, empirical research has been conducted across a
broad range of connected areas. The sheer range and diversity of these studies
in widely varying historical and geographical contexts cannot be enumerated
here. Confining ourselves to the fields of social and political theory, we note
how there have been path-breaking studies of the relationships between
states, as well as the latter’s interaction with a range of other social actors and
forces operating on the international stage.8
Research has also focussed on the
investigation of social and political struggles within states and organizations
at different levels of governance, not to mention the exploration of marginal-
ized identities, protest movements and the dynamics of public policy making
and implementation at a national and local level in a range of historical
contexts.9
Attention has also been paid to the uneven logics of globalization,
as well as to the role of the media in shaping discourses, language, and
identities.10
Nevertheless, despite these theoretical and empirical advances, there are
many who remain strongly critical or deeply sceptical about this tradition of
research. From the outset, theorists such as W.G. Runciman doubted the very
existence of structuralism, never mind the more exotic discourse of post-
structuralism, while as early as 1987 Anthony Giddens boldly declared both
to be ‘dead traditions of thought’, even though many of the more significant
poststructuralist contributions to social and political theory had yet to be
made (Giddens 1987: 195; Runciman 1970).11
Indeed, it is common
Introduction 5
Glynos-Intro.qxd 10/8/07 12:47 PM Page 5

for Marxists, positivists, critical realists and interpretivists, who agree about
precious little else, to forge a common front in dismissing the pretensions of
structuralists and poststructuralists as protagonists of a fashionable form
of linguistic idealism that is committed to nothing more than the ‘free play of
signifiers’ (Dreyfus and Rabinow 1982: 125).
However, even some who identify themselves as poststructuralist discourse
theorists, or are at least strongly sympathetic to its basic assumptions and
concepts, have raised important queries about the approach. Two such
criticisms – that we shall name the methodological and normative deficits
respectively – are especially pertinent to the arguments put forward in this
book. We need therefore to say a few words about the way we understand
these terms; the precise import of the allegations that have been made; and
how we propose to respond to these concerns.
The charge of a methodological deficit is raised in Jacob Torfing’s appraisal
of the current state of discourse theory, where he suggests among other things
that discourse theorists ‘must critically reflect upon the questions of method and
research strategy’ (Torfing 2005: 25). Here, however, it is important to stress
that the notion of a methodological deficit that underlies this challenge, as
well as the tasks it engenders, must be construed in the widest possible sense
of the term. To put it more precisely, following in the spirit of Marx, Weber
and Durkheim, while questions of method ought not to neglect the problems
that arise from the collection, analysis and status of empirical data, they most
definitely ought to focus on the full range of theoretical issues that arise in
the social sciences from the activities of describing, explaining, evaluating
and criticizing.
The upshot of this from our point of view is that methodological questions
necessarily touch upon the ontological and epistemological dimensions of any
social inquiry, as well as the specific techniques of data gathering and analysis
pertinent to a particular concrete case. This means that our response to this
methodological challenge must take up the full range of philosophical and
theoretical issues it necessarily implies. Indeed, as there are very few texts in
the poststructuralist tradition that tackle the question of method and the
nature of explanation in a sustained and philosophical way, we see our book
as helping to fill this gap.
If allegations of a methodological deficit question the ability of poststruc-
turalist discourse theory to explain and interpret, then the notion of a
normative deficit touches upon the issue of critique. This is neatly captured
by Simon Critchley’s assessment of Ernesto Laclau’s theory of hegemony, in
which he questions the capacity of poststructuralist discourse theory to evaluate
and transcend the existing order of things in the name of something new.
As he puts it
If the theory of hegemony is simply the description of a positively existing
state of affairs, then one risks emptying it of any critical function, that is,
of leaving open any space between things as they are and things as they
6 Introduction
Glynos-Intro.qxd 10/8/07 12:47 PM Page 6

may otherwise be. If the theory of hegemony is the description of a factual
state of affairs, then it risks identification and complicity with the
dislocatory logic of contemporary capitalist societies ...The problem
with Laclau’s discourse is that he makes noises of both sorts, both
descriptive and normative, without sufficiently clarifying what it is that
he is doing. This is what I mean by suggesting that there is the risk of
a kind of normative deficit in the theory of hegemony.
(Critchley 2004: 117)
Critchley’s critical comments epitomize a number of concerns raised about
the normative orientation and content of poststructuralism in general, and
discourse theory in particular.12
To express it more fully, if claims of a
methodological deficit raise questions about the explanatory capacity of
poststructuralist theory, then allegations of a normative deficit highlight
difficulties about the critical and reconstructive capacity of poststructuralist
theory. In short, does poststructuralism – and discourse theory in particular –
embody a new form of descriptivism or theoreticism? Again, our book develops
a set of middle-range categories that enable us to locate more precisely
the place and role of normative and ethical critique in poststructuralist
discourse theory.
In general, in responding to the methodological and critical challenges
outlined, it is important to stress that we avoid a number of possible though
ultimately unsatisfactory solutions. On the one hand, we resist the temptation
to offer a ‘method’ or ‘technique-driven’ solution to the alleged methodological
deficits, as this would blind us to the fact that any set of methods or
techniques is always relative to, and thus grounded upon, a particular
ontological stance (even if the latter is only implicitly evident in a series of
research practices or instruments).
On the other hand, we reject solutions that would involve a retreat into a
type of relativism or subjectivism where ‘anything goes’, because this response
would place no methodological constraint on the production and assessment
of putative explanations and critical evaluations. On the contrary, the whole
point of our book is to develop an ontological stance and a grammar of
concepts, together with a particular research ethos, which makes it possible
to construct and furnish answers to empirical problems that can withstand
charges of methodological arbitrariness, historical particularism, and idealism.
Indeed, in seeking to render our views on method more explicit, we hope to
provide the means to evaluate our approach and the studies it informs.
Finally, we reject the option of developing a comprehensive normative
framework, whether it takes the form of setting out the underlying principles
of social justice that ought to shape the basic structure of our institutional
arrangements, or whether it is predicated on articulating the fundamental
communicative and procedural pre-conditions for reaching a rational consensus
about common moral and political norms. As thinkers such as Connolly, Derrida
and Foucault have rightly insisted, the lure of elaborating a fully-fledged
Introduction 7
Glynos-Intro.qxd 10/8/07 12:47 PM Page 7

normative schema with which to evaluate and prescribe policies and practices
runs the risk of failing to engage with the singular instances of power,
domination and oppression that require careful empirical analysis, ethical
critique and political intervention. And yet we hope to show where and how
normative and ethical considerations can and ought to be brought to bear in
conducting concrete empirical research.
Logics of critical explanation
As is evident from the title of our book, our response to the challenges
levelled at positivism, hermeneutics, naturalism, and poststructuralism is to
elaborate an approach to social and political analysis that is captured by the
expression ‘logics of critical explanation’.13
By logics of critical explanation,
we refer to three senses of the term, each of which helps to flesh out our
approach in more detail. In descending order of generality, the notion first
picks out the ways in which processes of theory construction and explanation
are understood, involving very general considerations about problem construc-
tion, the selection of theoretical concepts, modes of reasoning in the sciences,
whether inductive, deductive or retroductive, and so on. Second, the concept
refers to a particular approach or ‘style of reasoning’ in the social sciences
(Hacking 1985), comprising the grammar of assumptions and concepts that
informs a particular approach to the social world: a way of formulating
problems, addressing them, and then evaluating the answers that have been
produced. For example, in developing our own approach we focus on positivist,
hermeneutical, critical realist, and poststructuralist styles of theorizing.
Finally, the concept is understood in a more substantive sense to constitute
the basic unit of explanation of our approach; logics in this sense contrast
with laws, self-interpretations, and mechanisms. Working within the field of
poststructuralism, our central aim in this regard is to construct an explana-
tory logic, together with the grammar of concepts and assumptions that serve
as its conditions of possibility, and to articulate a typology of basic logics –
social, political and fantasmatic – which can serve to characterize, explain and
criticize social phenomena.
The terms critical and explanation in ‘logics of critical explanation’ are more
often than not separated out in contemporary and modern conceptions of
philosophy and (social) science. Kant’s critical project drew a sharp line between
theoretical and practical reason – between the domains of knowledge and
morality – and positivists have widened the division between questions of fact
and explanation on the one hand, and questions of critique and normative eval-
uation on the other, by emphasizing the value neutrality of social scientific
inquiry. Weber also championed the cause of a ‘value-free’ social theory, though
for him this meant the ‘intrinsically simple demand that the investigator and
teacher should keep unconditionally separate the establishment of empirical
facts... and his own practical evaluations’ of them (Weber 1949: 11). Weber
thus distinguishes between the ‘value-relevance’ of the social sciences, in which
8 Introduction
Glynos-Intro.qxd 10/8/07 12:47 PM Page 8

he accepts that ‘judgements of interest’ are implicated in the choice of a particular
research object, and the necessary ‘value-freedom’ of the social scientist with
respect to the way she conducts her research and the way she uses the results of
her research. He thus rejects the belief that researchers should employ their
theories and explanations to support a specific political practice, project or
ideology (Hesse 1978: 9). At the heart of this more nuanced conception of the
relationship between values and social science is Weber’s wish to exclude
partisanship and overt political bias in the production and dissemination of
knowledge, though it stems in equal measure from his commitment to
naturalism in the sciences, and his ambivalent attitude toward law-like and
causal explanations in the natural sciences (Hesse 1978: 9–10).14
There have been a number of stern rebukes to these dominant under-
standings. For example, the first generation of Critical Theorists took their
target to be the complicities and partialities of ‘traditional theory’, which
they traced back to Descartes and the origins of modern philosophy, though it
was especially evident in the hegemony of nineteenth and twentieth century
positivism. Traditional theory in its various guises takes natural science and
especially mathematics to be its paradigm, thereby privileging deductive
and inductive reasoning, coupled with the desire to provide an as complete
and systematic picture of the world external to thought. However, in his
‘Traditional and Critical Theory’, Max Horkheimer stresses the way in which
the purveyors of traditional theory are submerged in particular social worlds
and practices, which they then serve to buttress by presenting the generation,
character and use of scientific knowledge as neutral and objective
(Horkheimer 1972: 190). Thus, what is overlooked is how its conception of
knowledge is integrally connected to the solving of problems within taken for
granted structures, especially the dominant forms of economic production.
Critical theory, by contrast, is conceived as a dialectical and engaged
practice that seeks to provide the means to challenge historically malleable
structures of domination and oppression in the name of a universal human
emancipation. Theoretical reasoning in this sense is a reflective rather than
purely technical and problem solving exercise that is concerned to highlight
and intensify the contradictions of a particular order, while simultaneously
proposing alternatives:
[T]he critical theory of society is, in its totality, the unfolding of a single
existential judgment. To put it in broad terms, the theory says that the
basic form of the historically given commodity economy on which modern
history rests contains in itself the internal and external tensions of the
modern era; it generates these tensions over and over again in an
increasingly heightened form; and after a period of progress, development
of human powers, and emancipation for the individual, after an enormous
extension of human control over nature, it finally hinders further
development and drives humanity into a new barbarism.
(Horkheimer 1972: 227)
Introduction 9
Glynos-Intro.qxd 10/8/07 12:47 PM Page 9

In his passionate critique of traditional theory, however, the early
Horkheimer relied upon a somewhat simplistic social theory, largely derived
from the Marxist tradition. He also remained committed to the power of
universal reason – though not in its purely Kantian manifestation since this
was seen to be too individualistic and disengaged – to expose the mismatch
between Enlightenment ideals and the ‘actually existing’ social practices and
institutions of modern capitalist societies.
Famously, however, perhaps notoriously, in the Dialectic of Enlightenment,
which Horkheimer co-authored with Theodor Adorno, this critique of
traditional theory and positivism mutated into a totalizing rage against reason
and the Enlightenment project itself (Bernstein 1991; Habermas 1987).
In this volte-face, Adorno and Horkheimer narrate the way in which the
Enlightenment ideals of rational demystification and disenchantment revert
back to a new form of ideological myth, thus ensuring the victory of an
instrumental reason that is complicit with the emerging ‘administered society’
(Adorno and Horkheimer 1973).
Also, in another strange twist, the second generation of critical theorists,
led mainly by Habermas, inverted this movement once again. In his early
work, Habermas launched an epistemological critique of science, especially in
its positivist incarnation, by showing the intrinsic connections between certain
sorts of knowledge and various forms of human interest (Habermas 1978).
But this was followed by a growing endorsement of the resources of universal
reason, now understood in communicative terms, to provide the means for
rationally agreed moral norms, so long as the appropriate procedures and
conditions could be secured. In this later conception, critique was linked to a
largely normative project that centred on the defence of universal rights and
procedures, as embodied in the modern liberal-democratic constitutional state
(Habermas 1996).
In the spirit of critical theory, though in a very different style, Roy Bhaskar’s
critical realism contests and re-inscribes the dominant distinctions and
oppositions. In his challenge both to positivists, who oppose fact and value,
and theoreticists, who split abstract theory and social practice, he develops a
practice of ‘explanatory critique’ in the human sciences that is explicitly
orientated towards ‘human emancipation’ (Bhaskar 1989: 102, 186). Indeed,
in his words, a suitably conceived and constructed social science makes the
connection between certain sorts of explanatory theory and the practices of
critique and evaluation ‘mandatory’ (Bhaskar 1989: 101, 105). Thus, while
Bhaskar does not posit an absolute identity between explanation and critique,
he argues that if certain theories (such as those informed by the philosophy of
critical realism) can identify false (that is, ideological) beliefs by providing
causal explanations of the sources of those beliefs, then we can and must move
immediately to a negative evaluation of the source of false beliefs, as well as
a positive evaluation of social action aimed at the latter’s challenge and
removal (Bhaskar 1989: 101–5). In short, well-founded explanatory theory has
intrinsic implications for critique and thus for human emancipation.
10 Introduction
Glynos-Intro.qxd 10/8/07 12:47 PM Page 10

However, while we accept Bhaskar’s attempt to articulate the practices of
explanation and critique in the development of a critical human science, we
take issue with the substance of his social theory, and we question the
particular ontological and epistemological framework within which it is
located. For our approach is predicated on the articulation of an alternative
social ontology that stresses the radical contingency and structural incompleteness
of all systems of social relations. In so doing, we draw heavily on what in
Lacanian theory is conceptualized as the disruptive presence of ‘the real’ in any
symbolic order, that is, a presence that marks the impossibility of any putative
fullness of being, whether at the level of structures, subjects or discourses
(Lacan 1978: 167; 1991a: 66; 2006: 324). Moreover, the effect of our ontological
framework is not only to destabilize the conditions upon which the standard
models of social science are grounded, but also to provide the conditions for
developing an alternative approach to social and political analysis that
inter alia concedes a central role to subjectivity (as distinct from subjectivism)
in characterizing, explaining and criticizing practices and regimes.
The centrality of this ontological starting-point requires us to say a little
more by way of its justification. Following Bhaskar and others, we begin by
assuming that any legitimate approach to social and political analysis requires
at least some ontological assumptions and commitments (see Connolly 2006;
Hay 2006). But the commitment to ontology can mean different things.
For some, the articulation of an ontological framework consists in providing
a kind of ‘furniture of the world’ that sets out the sorts of things, and their
respective properties, which we encounter in engaging with objects and other
subjects. However, in our view the importance of ontology is not just about
what sorts of things exist, but that they exist and how they exist. Indeed, of
capital importance in this regard is the fact that objects and subjects are
marked by an ‘essential instability’ that problematizes a simple listing of
their necessary intrinsic properties and causal capacities. Therefore, of greater
import for us is their contingency, historicity and precariousness. Indeed, in
terms of political analysis, this perspective enables us to highlight the constructed
and political character of social objectivity, and then to articulate a connected
series of concepts and logics that can help us to analyse social relations and
processes, while remaining faithful to our ontological commitments.
Argument and structure of the book
Working within this ontological framework, our overall argument is composed
of five inter-related elements: problematization, retroductive explanation,
logics, articulation and critique. Leaning on Foucault, we begin with the idea
of a problem-driven approach to social and political analysis. This involves
constructing theoretical and empirical objects of investigation, in which the
latter arise from pressing practical concerns of the present. Second, we
develop the idea of retroductive reasoning, which for us provides the grounds
for an appropriate form of explanation in the social sciences. Third, we
Introduction 11
Glynos-Intro.qxd 10/8/07 12:47 PM Page 11

introduce our concept of a logic, which is developed in opposition to
contextualized self-interpretations and causal mechanisms, and we elaborate
three kinds of logic that are linked together in a particular empirical context
to explain a problematized phenomenon.
This provides us with the appropriate content of a prospective explanation.
It is clear, however, that any putative explanans will comprise a plurality of
logics in a historically specific and complex set of social circumstances. This
means, fourth, that it is necessary to articulate these various types of logic
together in order to explain a constructed explanandum. Finally, but importantly,
the work of critique and evaluation in our approach does not precede – or follow
after – the practices of problematization, characterization, and explanation.
Instead, the task of social criticism is internally connected to them. This
involves the construction of a distinctive conception of ethics and normative
evaluation, in which the former is related to the radical contingency of social
relations, and the latter to historically specific relations of domination
and oppression.
The basic logic of our argument is reflected in the overall structure of the
book. In Chapter 1, we borrow from Aristotle, Peirce and Hanson to propose
the explicit adoption of retroductive reasoning as the paradigm for under-
standing the task of explanation in the social sciences. Retroduction is thus
opposed to the predominance of induction and deduction. We locate this retro-
ductive conception of explanation within an overarching logic of investigation
comprising the moments of problematization, retroductive explanation, persuasion
and intervention. In developing this argument, we first explore how retroductive
reasoning has been invoked to make sense of the practice of natural science.
Following Norwood Hanson, we show how retroduction can be used to
construct scientific theories by the process of positing hypotheses designed
to render recalcitrant phenomena more intelligible. However, in the natural
sciences explanation and testing are still construed in deductive-nomological
and hypothetico-deductive terms, and prediction is granted a constitutive role.
Indeed, this view makes possible a sharp distinction between the contexts of
discovery and justification (Popper 1980; Reichenbach 1938).
But things change when we move from the natural to the social sciences,
mainly because the deductive form of (exhaustive and predictive) testing and
explanation is problematized by our insistence upon the contextual particu-
larity of a putative explanation, as well as the latter’s always presupposed,
contestable framework of concepts and assumptions. Instead, we propose
different and more capacious criteria to justify and evaluate our explanations,
whose effect is to undermine positivism’s absolute separation between
the contexts of discovery and justification. In this view, the elements of
problematization, retroductive explanation, and persuasion are related,
though analytically distinguishable, parts of an interconnected whole. And
this means that retroductive reasoning can no longer be confined to the
context of discovery. Indeed, if explanation and testing cease to be linked in
an essential way to deduction and prediction, then the practice of explanation
12 Introduction
Glynos-Intro.qxd 10/8/07 12:47 PM Page 12

can be seen to intrude into the context of discovery. In short, then, the form
of explanation in social science ought to be understood in retroductive and
not deductive (or inductive) terms.
We then explore three possible contenders for filling out the content of our
explanatory form: contextualized self-interpretations, causal mechanisms,
and logics. Advocates of each of these contenders claim to – indeed for us
must – share two key insights. First, they accept the view that a social science
explanation must pass through and take seriously the self-interpretations of
the actors engaged in the practice under study. This is the ‘minimal’
hermeneutical insight that all three contenders take on board, thus distin-
guishing them from the positivist view, which is content to rely upon what
Charles Taylor calls ‘brute data identifications’ (Taylor 1985a: 28). The three
contenders also share a second insight: social science explanations are not
reducible to the self-interpretations of the social actors under study. In other
words, in explaining social and political phenomena, we can not rely exclu-
sively on what people say, or on their self-understandings, even though these
views must be taken into account in any legitimate social explanation.
The question then becomes one of fleshing out this ‘beyond’ of self-
interpretations.
Focussing on the work of Peter Winch, Charles Taylor, and Mark Bevir &
Rod Rhodes, the aim of Chapter 2 is to explore the hermeneutical account of
the social sciences, which makes contextualized self-interpretations the basic unit
of explanation. After initially demonstrating a convergence in the hermeneu-
tical critique of positivism and naturalism, we interrogate the explanatory
and critical capacity of hermeneutics more generally. Our claim here is that
while hermeneuticists highlight a number of deficiencies in the causal law
paradigm, they replace the latter’s subsumptive universalism with a descriptive
and normative particularism. This is problematic in our view because their
respective ontological frameworks fail to provide a satisfactory set of concepts
to characterize and explain social phenomena, but also because their approach
leads either to a diminishing of our critical capacity, which is evident in
Winch’s relativistic tendencies, or indeed to its precise opposite: an overbear-
ing normativism that results in a largely external critique of social practices,
generally predicated on a strong notion of rationality. The latter tendency is
evident in the writings of Taylor, as well as Bevir and Rhodes.
In Chapter 3, we explore the naturalist response to the failings of the
law-like model. Here our focus is on those who have substituted the role of
causal mechanisms for causal laws as the basic component of a legitimate
social science explanation. We concentrate mainly on the work of Jon Elster,
not simply because he is an exemplary advocate of the social mechanisms
approach, but also because his work shares important affinities with other
thinkers who seek to move beyond the causal law paradigm. For example,
Bhaskar’s naturalism appeals explicitly to the role of causal mechanisms in
explaining social phenomena, while the recent writings of Ian Shapiro,
Bob Jessop, and Colin Hay also make use of a critical realist conception of
Introduction 13
Glynos-Intro.qxd 10/8/07 12:47 PM Page 13

mechanisms and causal powers to develop an alternative account of social
science explanation.
But though Elster’s work mounts a powerful critique of the causal law
paradigm, especially the unrealistic demand to make the testing of predictions
a decisive criterion of explanation, we argue that the mechanisms approach
suffers in turn from a residual positivism that problematizes its distinctiveness
vis-à-vis the causal law approach. We do this by showing how the ideal of a
causal law continues to exert a powerful influence on Elster’s worldview,
arguing that this in turn leads to two problematic features of his approach.
These are what we term his psychologism and – perhaps somewhat counter-
intuitively – his idealism. In sum, we suggest that Elster does not so much
advance us through and beyond hermeneutics as return us to the positivism
of the causal law model. Moreover, we suggest that many aspects of our
argument are equally applicable to theorists such as Bhaskar, Shapiro and Hay.
Chapter 4 unfolds our alternative ontological framework for developing a
different approach to critical explanation in the social sciences. Influenced
principally by Heidegger, Lacan, and Laclau & Mouffe, but also drawing on
Foucault, Wittgenstein, and Derrida, we put forward an ‘ontology of lack’,
which is a negative ontology premised on the radical contingency of social
relations.15
Stated simply, we take this axiom to imply that any system or
structure of social relations is constitutively incomplete or lacking for a subject.
Consider in this regard Ernesto Laclau’s claim that every social identity is
always-already dislocated (Laclau 1990: 39). On the one hand, we take this to
be a strictly ontological understanding of dislocation, in which each and
every symbolic order is penetrated by an impossibility that has to be filled or
covered-over for it to constitute itself. The category of dislocation can also be
understood, however, in more ontical terms: moments in which the subject’s
mode of being is disrupted by an experience that cannot be symbolized within
and by the pre-existing means of discursive representation. From this
perspective, practices are governed by a dialectic defined by incomplete
structures on the one hand, and the collective acts of subjective identification
that sustain or change those incomplete structures on the other.16
Our fundamental ontological premise is then used to redescribe social
relations by stipulating different dimensions of social reality. The social
dimension captures those situations in which the radical contingency of
social relations has not been registered in the mode of public contestation,
whereas the political dimension refers to those situations in which subjects
responding to dislocatory events re-activate the contingent foundations of a
practice by publicly contesting and defending the norms of that practice.
On the other hand, the ideological and ethical dimensions of social reality
capture the way subjects are either complicit in concealing the radical
contingency of social relations (the ideological), or are attentive to its
constitutive character (the ethical).
In articulating this basic ontological standpoint, we take our principal
objects of investigation to be practices or regimes of practices, where our aim is
14 Introduction
Glynos-Intro.qxd 10/8/07 12:47 PM Page 14

to critically explain their transformation, stabilization, and maintenance.
Drawing on Heidegger, we claim that such an inquiry will always have an
ontical and an ontological impulse. Given the background ontological
framework laid out in Chapter 4, the following chapter introduces and
formalizes the category of a logic, which constitutes our basic unit of critical
explanation. In general terms, our conception of logic refers to the purposes,
rules and ontological presuppositions that render a practice or regime possible
and intelligible. An understanding of the logic of a practice aims, therefore,
not just to describe or characterize it, but also to capture the various conditions
that make that practice ‘work’ or ‘tick’.
More concretely, in the realm of social science explanation, we delineate
three kinds of logic – social, political and fantasmatic – that speak to different
dimensions of social reality. Closely associated with the social dimension,
social logics enable us to characterize practices in a particular social domain
(for example the practice of modularization within universities), or an entire
regime of practices (for instance Thatcherism, apartheid, or the audit regime).
Political logics provide the means to explore the conditions of possibility and
vulnerability of social practices and regimes by focusing on the latter’s
contestation and institution. Here we invoke Laclau and Mouffe’s logics
of equivalence and difference to investigate the way in which the traces of
radical contingency associated with the original institution of practices and
regimes can in certain circumstances be reactivated by subjects, thus enabling
them to construct new meanings, practices and identities. Finally, if political
logics are most closely associated with the political dimension of social
relations, fantasmatic logics are closely linked to the ideological dimension. In
this regard, the logic of fantasy, which is predicated on the Lacanian category
of enjoyment ( jouissance), shows how subjects are rendered complicit in
concealing or covering over the radical contingency of social relations.
However, insofar as fantasmatic logics are subverted we also have a means of
accounting for conditions under which the ethical dimension of social reality
can be foregrounded.
Since the social, political, ideological and ethical dimensions of social reality
are always to some degree present in any particular practice or regime, each
of the three logics has a role to play in articulating a complete explanatory
account. This means that our overriding aim in Chapter 5 is to establish the
distinctive features and status of each of these logics by relating them to our
overall social ontology. At the same time, we are able to distinguish our
approach from those that appeal to contextualized self-interpretations and
causal mechanisms. While in our view logics are subject-dependent, in the
sense that our explanations require a passage through a subject’s contextualized
self-interpretation (the hermeneutical constraint), they also require some-
thing that transcends them. However, as we argue in Chapter 3, this ‘something
more’ is not adequately addressed by theorists such as Elster, Bhaskar and
Shapiro. We seek instead to develop this ‘something more’ in a way that
remains more faithful to the hermeneutical insight.
Introduction 15
Glynos-Intro.qxd 10/8/07 12:47 PM Page 15

While much of our discussion in Chapters 4 and 5 focuses on the explanation
of problematized social phenomena, we also stress that our approach carries
important critical implications. In particular, we argue that political and
fantasmatic logics not only provide the means to explain the emergence and
reproduction of practices and regimes, but they also highlight the contingency
of the latter’s institution and grip. Political and fantasmatic logics thus
furnish the means for the ethical critique and normative evaluation of practices
and regimes. In rendering these ideas more explicit, we draw upon
William Connolly’s notion of ontopolitical interpretation (Connolly 1995) to
show how different logics are linked together as part of an ethicopolitical
practice of critical explanation. The theme of critique in social science
explanation is returned to in Chapter 6, where we emphasize both the
normative and ethical aspects of this practice.
However, the main thrust of Chapter 6 is to address a number of method-
ological and epistemological issues that are raised, but not fully addressed, in
previous chapters. These include questions pertaining to the epistemological
and methodological status of case studies within our approach; the role of
comparative research in deepening, modifying and ‘testing’ our critical
explanations; the conceptualization of the relationship between theoretical cat-
egories and empirical analysis, especially the way in which different theoretical
and empirical elements come together to form an explanatory narrative; as
well as the implications of our approach for the way we should understand
hypothesis-generation, validation, generalisation and theory-building.
We respond to these questions by condensing many of the issues of social
science explanation into what we call the problem of subsumption. The latter
denotes an understanding of the relationship between concepts and objects as
external to each other, in which objects are gathered under concepts without
the object or the concept undergoing any modification during the process of
subsumption. But in critical opposition to subsumption, we develop the con-
cept and practice of articulation in order to characterize the way we ought to
understand both social science research practice and many aspects of the gen-
eral logic of critical explanation. A major part of this discussion centres on
our preferred understanding of the role of case studies, comparative research
and generalization in the social sciences. We also connect our concept of artic-
ulation to the practice of judgement, as it is understood by thinkers such as
Kant and Wittgenstein, showing how this conceptual linkage enables us to
address issues concerning the application of abstract theoretical concepts and
logics to empirical objects, or how to incorporate into our approach theoret-
ical concepts and empirical generalizations derived from other traditions of
thought.
In discussing these epistemological and methodological issues, we illustrate
our approach by problematizing the recent reforms and changing practices of
the UK’s higher education system, which involve the introduction, sedimen-
tation, and maintenance of a new audit regime. Though this by no means
constitutes a case study, we do spend some time outlining the background
16 Introduction
Glynos-Intro.qxd 10/8/07 12:47 PM Page 16

context to this regime so as to better illustrate the overall contours of our
approach. What in other words would the relevant social, political, and
fantasmatic logics look like or imply in such a case? Drawing on this
discussion, we also highlight the implications of our ‘method’ of articulatory
practice for issues of critique. More specifically, with reference to our study of
the higher education audit regime, we introduce the notion of counter-logics
to flesh out our conception of critical explanation. Further, by developing
themes introduced in Chapters 4 and 5, we show how our account of logics
enables us to situate and relate the tasks of ethical and normative critique in
poststructuralist discourse theory. But before we can develop these implications
we need to begin by setting out the general form of critical explanation in
social and political analysis.
Introduction 17
Glynos-Intro.qxd 10/8/07 12:47 PM Page 17

The ultimate goal of a positive science is the development of a ‘theory’ or
‘hypothesis’ that yields valid and meaningful (i.e., not truistic) predictions
about phenomena not yet observed.
(Friedman 1953: 7)
We have reason to wonder if human activity will ever lend itself to
fundamental analysis by truly scientific methods – whether there is a mystery
about man that makes his actions largely unpredictable.
(Duverger 1972: 6)
Faith in mathematical and statistical methods in the social sciences is
surprisingly tenacious in the face of repeated failures to meet positivist ideals
of explanation and prediction.1
This chapter makes the case for a post-positivist
paradigm of explanation by contesting the causal law paradigm, which
privileges prediction as a constitutive element of the explanatory process.
Of course, this aim is not new (see Steinmetz 2005). Nevertheless, in view of
the tenacity of the positivist impulse in social science to survive in different
guises, we need to continue to explore new ways of challenging and engaging
with it, as well as new ways of conceptualizing alternatives. As Stephen
White points out in the context of political studies, ‘[a]lmost all political
theorists and scientists affirm some notion of a post-positivist model of inquiry.
But when pushed to explain exactly what is meant by that, both at the
theoretical-philosophical level, as well as at the level of concrete research,
most of us become relatively inarticulate quite quickly’ (White in Topper et al.
2006: 734).
In responding to White’s challenge, our book sets out to recast the perennial
debates between positivism and its ‘others’ from a poststructuralist point of
view. In this chapter we claim that the default tendency to rely on deductive
(and inductive) reasoning, the hypothetico-deductive method of scientific
investigation, and the covering-law model of explanation, is a product of the
hegemonic grip of a particular, though admittedly powerful, conception of
the natural sciences – what we call the causal law paradigm.2
Our main critical
argument is that it is problematic to model social processes on natural
1 Retroduction
Glynos-01.qxd 10/8/07 12:51 PM Page 18

processes in this way – whether as universal laws, causal generalizations, or
robust empirical correlations – because it leads to rather narrow conceptions
of testing and explanation, in which the element of prediction is elevated at
the expense of contextual and ontological factors. In short, our target is the
law-like conception of explanation and testing that the causal law paradigm
elevates to the status of an ideal.
As against inductive and deductive modes of reasoning, we argue more
positively that retroductive reasoning provides us with a general form or logic
of explanation in the social sciences. Although Aristotle has been credited
with its original identification, retroduction has been insightfully developed
and applied in the philosophy of science by Charles Sanders Peirce, Norwood
Hanson, Roy Bhaskar and others. However, while we draw upon these
scholars, it is only by being sensitive to the distinctive ontologies underpinning
the natural and social worlds that we can fully reap the potential of retroduction
for the social sciences. In so doing, we rework the sharp distinction drawn in
positivist images of social science practice between what we call, following
Reichenbach, the ‘context of discovery’ and the ‘context of justification’
(Reichenbach 1938). Despite Reichenbach’s staunch defence of induction as
a means of distinguishing science from non-science, it is widely accepted that
Popper’s appeal to falsificationism as an alternative criterion of demarcation
also relied on this distinction (Popper 1980). And though the reworking of
this distinction has been a staple of the history and philosophy of science ever
since its introduction,3
we do so in order to challenge the positivist conception
of causal explanation in the social sciences in a particular way.4
More
specifically, we challenge the compartmentalizing tendencies of positivist
social science investigation – a logic of scientific discovery followed by
exhaustive empirical testing and explanation – and propose instead one over-
arching logic of investigation comprising three interlocking moments: the
problematization of empirical phenomena; the retroductive explanation of these
phenomena; and the persuasion of – and intervention into – the relevant
community and practices of scholars and lay-actors.
These ideas are developed in three steps. First, we introduce the concept of
retroduction as it has been invoked in the philosophy of natural science.
Although in our view retroductive reasoning is present in all forms of
scientific practice, we follow Norwood Hanson in suggesting that it is either
left implicit in the practices of scientists and/or confined to a particular realm
of scientific activity, namely, the activity of conceiving and proposing
hypotheses (Hanson 1961: 71). Second, we show how retroduction can and
should be made relevant to social science practice and explanation. This
involves indicating the way it has already been discussed in the philosophy of
social science by theorists such as Bhaskar and Derek Sayer, as well as
developing our own argument about how its relation to social science practice
and explanation should be understood. Our argument specifies the way in
which the ontological shift from the natural to the social world results in our
abandoning the positivist understanding of the distinction between the contexts
Retroduction 19
Glynos-01.qxd 10/8/07 12:51 PM Page 19

of discovery and justification. In a third step, we explore the consequences of
this insight for thinking about practices of theory construction and critical
explanation in the human sciences.
The problem of prediction in social science
One reason to be sceptical about transforming prediction into a constitutive
feature of social science explanation might be the dramatic failures of social
scientists to anticipate major economic and political events, whether the
South East Asian financial crisis of 1997, the fall of the Berlin wall in 1989,
or the relatively peaceful and stable transition from Apartheid to democracy
in South Africa. Of course, objections can be raised against this position.
For instance even the natural sciences are often unable to predict many
individual phenomena (such as weather events) with satisfactory precision.
And yet the question still remains as to whether it is plausible or desirable to
hold up prediction as a constitutive feature of social science explanation.
From this perspective, it appears that social scientists do not fare well, even
in their very own (positivist) terms, whether in economics or political science
(see Tetlock 2005). As some scholars have put it, again in relation to the
study of politics, ‘the program of identifying simple general laws concerning
political structures and processes has so far yielded meagre results’, its
strength lying ‘in its identification of empirical regularities to be explained,
not in its provision or verification of explanations’ (Tilly and Goodin
2006: 20).5
The privileging of prediction is often justified by the need to have a good
basis upon which to retain or reject a model. ‘If two different models produce
different predictions then we know they are competing. But if we are not
sure what the predictions are from two models we do not even know if they
are competing. They might [even] be the same model’ (Dowding 2001: 93).
And Rebecca Morton is cited approvingly in this context: ‘Empirical analysis
that never builds towards an explicit set of assumptions and predictions about
the real world is no better than pure description’ (Morton 1999: 44).
But in hypostatizing prediction in this way, social scientists run the risk of
prejudging, and thus severely restricting the scope and nature of social science
research, in the style of ‘if a hammer is all you’ve got, then sooner or later
everything starts to look like a nail’. As Ian Shapiro points out,
making a fetish of prediction can undermine problem-driven research via
wag-the-dog scenarios in which we elect to study phenomena because
they seem to admit the possibility of prediction rather than because we
have independent reasons for thinking it worthwhile to study them. This
is what I mean by method-drivenness...In principle it sounds right to
say ‘let’s test the model against the data’. In reality, there are few
uncontroversial data sets in political science.
(Shapiro 2002: 609)6
20 Retroduction
Glynos-01.qxd 10/8/07 12:51 PM Page 20

In our view, the tendency to hypostatize prediction reflects the desire among
social scientists to imitate the ideals of what they take to be modern natural
science, especially physics, as well as to adopt the ontological commitments
that sustain these ideals.7
As Dowding puts it, ‘[t]he interpretations we
privilege in the end are the ones that provide the best predictions in precisely
the same manner as our interpretations of the patterns we ascribe to the physical
world’ (Dowding 2004: 140; emphasis added). This view reflects the
dominance of the hypothetico-deductive method of science, and its associated
conception of causal explanation, which is taken as a kind of formalization of
natural science practice. It is to the basic features of this method and its
particular style of reasoning that we now turn.
As Hubert Dreyfus has argued, this model of natural science is regarded by
many as the crowning glory in the development of ‘ideal theory’, and it can
be traced back to Socrates and Plato through to Descartes and Kant in the
modern period (Dreyfus 1986: 11). It consists of six essential characteristics:
explicitness (a theory must be fully spelled out, and not based on intuition and
interpretation); universality (a theory should hold true for all places at all times);
abstractedness (a theory must not refer to particular examples); discreteness
(a theory must be stated in terms of context-free elements that are not
reducible to human interests, traditions, institutions, and so forth); and
systematicity (a theory must be a whole in the sense that decontextualized
elements – attributes, features, factors, etc – are related to each other by
rules or laws) (Dreyfus 1986: 12). Finally, it was left to the modern natural
sciences – the theories propounded by Copernicus, Galileo, Kepler, Newton,
and so forth – to add the sixth characteristic, namely, predictive power and exhaus-
tiveness. In this regard, theory must provide a complete description of the
domain investigated, that is, it must specify all types of changes affecting
the domain, as well as specifying their effects. Importantly, as Dreyfus puts it,
‘this completeness permits precise prediction’ (Dreyfus 1986: 12).
This idealized model of science, which of course does not have to be realized
to function as an ideal, is usually instantiated in the hypothetico-deductive
method and the corresponding ‘covering law’ or ‘deductive-nomological’
model of explanation. In both, the idea of prediction is allotted a constitutive
role vis-à-vis theory-building and explanation. Simplifying Popper’s classic
formulation of the hypothetico-deductive method, the potential falsifiability
of a proposed hypothesis – its capacity to be disproved by observed facts – serves
as a demarcation criterion for a scientific statement. Theories are corroborated
or falsified by deducing empirical statements or predictions, and then testing
these against observational evidence.8
Deduction, therefore, assumes a
prominent role within the context of justification, that is, with respect to the
testing, verification, falsification, formalization, and presentation of theories.9
The allied task of explaining facts, following Carl Hempel’s widely accepted
account, is ‘deductive-nomological’ in form, in which an explanandum is
deduced from a set of premises that includes a universal law. More precisely,
laws comprise law-like statements from which, given an appropriate set of
Retroduction 21
Glynos-01.qxd 10/8/07 12:51 PM Page 21

initial conditions, the fact that has to be explained can be logically derived
(Elster 1983a: 26). And importantly, from our point of view, it is here that
the boundary between explanation and prediction tends to blur. This is
because explanation and prediction are linked together in a deductive framework,
their difference amounting only to a difference in temporal perspective:
‘Explaining x is predicting x after it has actually happened...Predicting x is
explaining it before it has actually happened’ (Hanson 1972: 41). Or, to put
it in Popper’s terms, ‘the use of a theory for the purposes of predicting some
event is just another aspect of its use for the purpose of explaining such an
event’ (Popper 1966: 262–3).
To use Hempel’s stylised example, if we wish to employ the deductive-
nomological model to explain the fact that our car radiator iced up on a cold
night, we require a set of initial conditions – the liquid in our car radiator
was water; the radiator did not leak; and the temperature fell below zero
degrees Celsius – coupled with a law-like statement to the effect that water
freezes at zero degrees Celsius. When these elements are combined they
generate a particular account (explanans) which explains deductively the event
or fact (explanandum). In addition, explanation enables us to predict future
(or retrodict past) occurrences, so that we can take precautions and insert
anti-freeze in our radiator during the winter (Hempel 1965: 232).
More formally, the deductive-nomological model can be expressed in the
following way:
1 Law or theories
2 Initial conditions
3 Event or fact explained and/or predicted
And the above example can be displayed as in Table 1.
The deductive-nomological model of explanation and prediction can also be
used to express ‘Duverger’s law’ (Duverger 1959) – the claim that under certain
conditions the simple majority electoral system favours the two-party system –
one of the few ‘laws’ that have been confidently asserted by political scientists
(see Connolly 1981: 9–18). Stated simply, it can take the form of Table 2.
It is clear from Table 2 that Duverger’s law enables political scientists to
explain the relationships between certain sorts of electoral and party systems,
as well as to predict future relationships – or retrodict past relationships – thus
making the practice of institutional design more reflective and efficient.
22 Retroduction
Table 1 The deductive-nomological model: a natural science illustration
Explanans Law Water freezes at 0 degrees
Initial conditions Liquid in car is water; the radiator did
not leak; temp falls below 0 degrees
Explanandum Event/process Therefore radiator freezes
Glynos-01.qxd 10/8/07 12:51 PM Page 22

We shall return to Duverger’s law later in the chapter. For now, it suffices
to note that the structure of subsuming an event under law-like statements
parallels the role of prediction in the hypothetico-deductive model. In other
words, the explanandum is explained by facts which, had they been known in
advance, would have allowed us to predict it (Lessnoff 1974: 22). The grip of
the hypothetico-deductive and deductive-nomological pictures of science,
coupled with the obvious success of the natural sciences, explains why it
appears natural for many social scientists to link prediction and explanation
so rigidly. However, these claims only appear natural if one accepts the unity
of method in the sciences – and the hypothetico-deductive method more
precisely – while remaining unperturbed by the failures and anomalies of the
positivist project in the social sciences.
Of course, these observations do not by themselves invalidate positivism.
Even the most dramatic failures of prediction cannot conclusively falsify the
central tenets of positivism.10
Indeed, far from falsifying its research programme,
these failures can serve to demonstrate that social scientists are not positivist
enough, and that they should embrace the spread of more sophisticated
mathematical and quantitative techniques tout court. Construed as empirical
deficiencies, the failure of positivism is thus understood in epistemological
terms, leaving intact the ideals and ontological assumptions borrowed from
the natural sciences.11
But equally, this does not mean that the questioning of predictive capacity
as an essential component of explanation ought to imply an ontological
structure that is intrinsically undetermined or infinitely malleable – a pure
flux of random events and processes. After all, as has been repeatedly shown
in the domain of complex adaptive systems and chaos theory, phenomena can
be both fully determined and unpredictable (Schroeder 1991: 167; see also
Prigogine 1996 and Stengers 1997). The key is to locate the right level at which
invariance is operative in the world, and to ask how best to characterize it. As far
as the social world is concerned, we argue that though this invariance may be
historically specific, it is a structure or pattern shaped in significant ways by
the (ontological) fact that we are ‘meaning-producing’ or ‘self-interpreting’
animals (see Taylor 1985a: Ch. 2). In this chapter, then, we draw initially and
minimally upon hermeneutical thinking to question the law-like model of
explanation and prediction in the social sciences, leaving it until Chapter 2
Retroduction 23
Table 2 The deductive-nomological model: a social science illustration
Explanans Law The simple majority electoral
system favours the two-party system
Initial conditions Legalization of party organizations; secret
ballot; inclusion of party affiliation on
ballots; diffusion of ethnic/class
membership across electoral districts; etc.
Explanandum Event/process Therefore x party system is two-party
Glynos-01.qxd 10/8/07 12:51 PM Page 23

to explore this tradition more fully and to engage with it more critically.
However, before we explore this ‘minimal’ hermeneutical program, we wish
to consider at least one productive affinity between social science and natural
science, which will be crucial to our later argument.
Retroduction and the philosophy of science
Following Reichenbach and Popper, we have shown how explanation in the
natural sciences is closely associated with what we take to be the context of
justification, suggesting there are at least some reasons, though by no means
conclusive ones, for being sceptical about importing natural science conceptions
of testing, prediction, and explanation into the social sciences. In this section,
we are more optimistic about drawing out some parallels between the two
sciences, which centre on a distinct form of reasoning pertaining to the
context of discovery in the natural sciences. This form of reasoning has been
called retroduction or abduction. Given certain facts or anomalies (conclusions)
retroductive reasoning describes the way plausible hypotheses are produced
(our search for premises). It is prima facie appealing from a social science
perspective because it designates a backward-looking modal form of inference
with which many social scientists are very familiar. For example, it is not
unusual for social scientists to take the resounding hegemonic success of
Thatcherism as the given anomaly, and then to proceed backwards to furnish
an account of how and why this was so (Gamble 1994; Hall 1988).
Norwood Hanson is one of a number of philosophers of science who has
suggested that this form of reasoning best captures the nature, practice, and
development of natural science, especially the practice of theory-construction.
He invokes retroductive reasoning to suggest that Kepler, for example, ‘did
not begin with the hypothesis that Mars’s orbit was elliptical and then deduce
statements confirmed by Brahe’s observations. These latter observations were
given, and they set the problem – they were Johannes Kepler’s starting point’
(Hanson 1961: 72). After a detailed account of the way Kepler arrived at his
‘ellipse hypothesis’ Hanson concludes that Kepler
wrote De Motibus Stellae Martis in order to set out his reasons for suggesting
the ellipse. These were not deductive reasons; he was working from
explicanda to explicans. But neither were they inductive – not, at least, in
any form advocated by the empiricists, statisticians and probability
theorists who have written on induction.
(Hanson 1961: 85)
As Hanson notes, alongside inductive and deductive types of inference,
Aristotle lists a third – translated by Peirce as ‘abductive’ or ‘retroductive’
reasoning – which consists ‘in studying facts and devising a theory to explain
them’ (Peirce 1934: 145).
24 Retroduction
Glynos-01.qxd 10/8/07 12:51 PM Page 24

In more formal terms, Peirce offers the following definitions of these three
forms of reasoning. He begins with deduction, which
is that mode of reasoning which examines the state of things asserted in
the premises, forms a diagram [whether geometric, algebraic, or otherwise]
of that state of things, perceives in the parts of that diagram relations not
explicitly mentioned in the premises, satisfies itself by mental experiments
upon the diagram that these relations would always subsist, or at least
would do so in a certain proportion of cases, and concludes their necessary,
or probable, truth.
(Peirce 1960: 28)
Induction, by contrast,
is that mode of reasoning which adopts a conclusion as approximate,
because it results from a method of inference which must generally lead
to the truth in the long run. For example, a ship enters port laden with
coffee. I go aboard and sample the coffee ...I conclude by induction that
the whole cargo has approximately the same value per bean as the hundred
beans of my sample.
(Peirce 1960: 28)
Retroduction, finally,
is the provisional adoption of a hypothesis...For example, all the
operations of chemistry fail to decompose hydrogen, lithium, glucinum,
boron, carbon, nitrogen, oxygen, fluorine, sodium...We provisionally
suppose these bodies to be simple; for if not, similar experimentation will
detect their compound nature, if it can be detected at all.
(Peirce 1960: 29)12
The conceptual irreducibility of retroduction, and thus its integrity, is
important for Hanson because it provides the key to a better account of the
practice of theory-building in the natural sciences (Hanson 1961). Not only
does it furnish a corrective to the way scientists understand their own
practice, but it also supplies a corrective to the way scientific practices are
understood by philosophers and historians of science. ‘Physicists rarely find
laws by enumerating and summarising observables’, argues Hanson, nor do
they ‘start from hypotheses’; instead, ‘they start from data’ (Hanson 1961: 70).
Indeed, as he puts it, the ‘struggle for intelligibility (pattern, organization)
in natural philosophy has never been portrayed in inductive or H-D
[hypothetico-deductive] accounts’ (Hanson 1961: 87).
Retroduction is at one with induction in suggesting that the vector of
reasoning points from the data to the laws. However, as Hanson insists,
Retroduction 25
Glynos-01.qxd 10/8/07 12:51 PM Page 25

inductive and deductive reasoning cannot originate any theories or laws
whatever. From an inductive point of view, this is because theories are simply
summarized projections of these data, while from a deductive perspective
they are derived from a law or an axiom. Retroduction by contrast moves
from data to hypothesis to laws.13
In short, though more open-ended than
deduction and induction, retroductive reasoning is still a legitimate type of
inference. While deductive reasoning purports to prove what is the case, and
inductive reasoning purports to approximate what is the case, retroductive
reasoning conjectures what is the case (Peirce in Hanson 1961: 85). As Peirce
puts it, retroduction is a ‘logical inference, asserting its conclusion only
problematically, or conjecturally, it is true, but nevertheless having a perfectly
definite logical form’ (Peirce in Hanson 1961: 86).
Hanson is all too aware of the temptation to dismiss retroduction on
account of its apparent unwieldiness, abandoning it to the unfathomable
whims of circumstance. From this point of view, at least induction has the
decency to attempt to give some non-arbitrary content to the discovery
process. But Hanson vigorously disputes the allegation that the generation of
hypotheses is
so often affected by intuition, insight, hunches, or other imponderables
as biographers or scientists suggest. Disciples of the Hypothetico-
Deductive account often dismiss the dawning of a hypothesis as being of
psychological interest only, or else claim it to be the province solely of
genius and not of logic. They are wrong...To form the idea of acceleration
or of universal gravitation does require genius: nothing less than a Galileo
or a Newton. But that cannot mean that the reflexions leading to these
ideas are unreasonable or a-reasonable. Here resides the continuity in
physical explanation from the earliest to the present times.
(Hanson 1961: 72; see also Hanson 1972: 66–7)
Indeed, the single most important criterion for admitting a hypothesis,
however tentatively, is that it accounts for the phenomenon or problem at
stake. In this sense, retroductive reasoning takes a three-fold form. To begin
with, a surprising, anomalous, or wondrous phenomenon is observed (P). This
phenomenon ‘would be explicable as a matter of course’ if a hypothesis (H)
were true, and so there is good reason to think that H is true (Hanson 1961: 86).
In other words, the hypothesis is not inferred until its content is already
present in the explanation of P. This contrasts with inductive accounts that
‘expect H to emerge from repetitions of P’ and with H-D [hypothetico-
deductive] accounts which ‘make P emerge from some unaccounted-for
creation of H as a “higher-level hypothesis”’ (Hanson 1961: 86).14
It is important to stress that although Peirce and Hanson see no conceptually
necessary connection between a phenomenon (P) and a hypothesis (H),
this does not mean that there are no ‘clear criteria governing what count as
acceptable putative explanans, whether or not we choose to dignify its
26 Retroduction
Glynos-01.qxd 10/8/07 12:51 PM Page 26